We address a continuous space search problem, where the searcher and the mobile target are moving on a topographic terrain, and the goal of the searcher is to locate the target in minimum expected time. The detection occurs when the target is located in the searcher's observed area. We assume that the searcher's observed area is bounded by topographic obstacles and changes in time along with the searcher's trajectory. In the report we provide a general algorithm of search, which is applicable to the search for both static and mobile targets. It follows a two-step solution: the first step is a terrain analysis, and at the second step the searcher's path is determined. For all simulated topographies the suggested algorithm converges and provides near-optimal solutions for the search of both static and moving targets even in the cases, where standard methods fail to provide solutions in polynomial time.
Video streaming is now responsible for the majority of Internet traffic and is expected to keep growing over the coming years. Dynamic Adaptive Streaming over HTTP (DASH) [1] is an ISO/IEC MPEG multi-quality layer streaming solution that is designed to enable interoperability between servers and clients of different vendors. In the DASH protocol, the client-side player is assumed to have Adaptation Logic (AL). The AL evaluates the various video representation segments available on the server and chooses the most suitable segments balancing between video quality and switching time. Note that dynamic adaptation is necessary due to the fact that the network bandwidth (e.g. cellular network) and the user's buffer are not stable and have a high influence on re-buffering. However, to date, none of the research considers multicast conditions and therefore, there is no AL specifically designed to support multicast at the client side. In this paper, we present the Harmonic Mean Adaptive Logic (HMAL) which is a buffer sensitive adaptation logic that first calculates how many segments exist in the buffer and then estimates the channel bandwidth using the harmonic mean of the previous n samples. The HMAL is designed to support multicast networks by reducing the weights of hight quality segments in the bandwidth estimation and give more weight to lower quality segments. Compared to the multicast versions of well known ALs, the simulation results showed that HMAL has the best bandwidth estimation, the lowest number of re-buffering events, and the highest buffer efficiency.
In this paper, we will show two methods of digital closed loop envelope tracking systems for high bandwidth communication systems, tested on wifi 80 MHz signals (11ac) in high level simulation platform (matlab & simulink). We will present the challenges of high frequency digital DC2DC converter for envelope tracking (ET) as load modulation wide digital controller loop and noise and saturation in the loop, and multilevel ET based on several high BW digital low dropout (DLDO) voltage regulators, and examine the differences in terms of the challenges and coping mechanisms each one holds. We will show that although the existing highly non-linear PA in our system, both architectures provide the solution in terms of band edge (BE) mask limitation, with efficiency of ~80% for the ET full driver. All components in both architectures are fully digital and synthesizable, which has a big advantage compared to analog options in efficiency noise and size, and of course assures robustness and flexibility.
This experiment was designed to see if information related to linguistic characteristics of read text can be deduced from fMRI data via machine learning techniques. Individuals were scanned while reading text the size of words in loud reading. Three experiments were performed corresponding to different degrees of grammatical complexity that is performed during loud reading: (1) words and pseudo-words were presented to subjects; (2) words with diacritical marking and words without diacritical markings were presented to subjects; (3) Hebrew words with Hebrew root and Hebrew words without Hebrew root were presented to subjects. The working hypothesis was that the more complex the needed grammatical processing needed, the more difficult it should be to perform the classification at the level of temporal and spatial resolution given by an fMRI signal. We were able to accomplish the first task completely. The second and third task did not succeed when all the data is used simultaneously. However, the third task was successful when training and testing was done within a continuous scanning run. (The experimental protocol did not allow this for the second task.) This does establish that complex linguistic information is decodable from fMRI scans. On the other hand, the need to restrict to the intra-run situation indicates that additional work is needed to compensate for distortions introduced between scanning runs.
As the quantity of visual information sources increases, the need to develop sensors that can automatically alert the user to exceptional events is being emphasized. This feature incorporates the ability to detect and classify the targets they “see”. Achieving this goal will dramatically improve the efficiency of CCTV-based security systems, improve search/retrieval engines, increase the autonomy of robotic systems and will contribute in many other areas of life. The performance of the human visual system and its robustness to image degradations still surpasses the best computer vision systems. Remarkable in particular, is the human brain high accuracy in ultra rapid object categorization tasks. Recent studies shows that the mechanism behind the recognition process includes predictions based on prior knowledge about the world. These predictions enable rapid generation of hypothesis that biases the outcome of the recognition process in situation of uncertainty. In this work, we implemented the concepts behind this top-down prediction mechanism of the human visual system. This work focus on the orbitofrontal cortex (OFC) role in the prediction process, which appears to be attuned to the associative content of visual information and to facilitate recognition of sensory inputs via predictive feedback to sensory cortices. Specifically, coarse representations reach the OFC, which generate “initial guesses” regarding the targets identity. These predictions are projected to the inferior temporal cortex, which facilitate perception and select the most likely interpretations. We show that imitating this mechanism can potentially create more robust target recognition models than exist today.
This work presents an analysis of the behavior of an electromagnetic field near the common edge of two resistive half-planes with different surface impedances. Unlike a single resistive half-plane, in the case of the impedance junction, both transverse electric and magnetic field's components simultaneously contain a logarithmic singularity. It was shown that a surface current density has a finite jump that is proportional to the difference between the inverse impedances.
One of the important tasks of a humanoid-robot auditory system is speaker localization. It is used for the construction of the surrounding acoustic scene and as an input for additional processing methods. Localization is usually required to operate indoors under high reverberation levels. Recently, an algorithm for speaker localization under these conditions was proposed. The algorithm uses a spherical microphone array and the processing is performed in the spherical harmonics domain, requiring a relatively large number of microphones to efficiently cover the entire frequency range of speech. However, the number of microphones in the auditory system of a humanoid robot is usually limited. The current paper proposes an improvement of the previously published algorithm. The improvement aims to overcome the frequency limitations imposed by the insufficient number of microphones. The improvement is achieved by using a novel space-domain distance algorithm that does not requires the transformation to the spherical harmonics domain, thereby avoiding the frequency range limitations. A numerical study shows two important results. The first is that, using the improved algorithm, the operation frequency range can be significantly extended. The second important result is related to the fact that higher frequencies contain more detailed information about the surrounding sound field. Hence, the additional higher frequencies lead to improved localization accuracy.
This paper concerns the problem of path planning for a team of autonomous mobile robots, tasked with the formation of a relay network for providing communications between a stationary object and a remote station (controller). The trajectories of the robots should be determined so as to optimize their energy required for communication and motion. The robots are assumed to transmit while in motion. The problem is formulated in the setting of optimal control, and solved by the application of an efficient algorithm recently developed by the authors. The algorithm is based on explicit minimization of the Hamiltonian at each step thereby providing an effective, easily-computable descent direction. Following the problem formulation the paper presents the algorithm and its convergence properties, demonstrates it on application examples, and discusses various directions for future research.
With scaling of process technologies and worsening of process variations, embedded memories are susceptible to a large number of failure mechanisms making it hard to achieve high yield. In this paper, by bringing together architecture and circuit-level exploration tools, we analyse the impact of process variations on static random access memory (SRAM) cell stability and determine the impact of SRAM failures on memory functional yield. We then detail the importance of repair mechanisms such as error correcting codes (ECC) and redundancy on improving yield subject to constraints set on power and area. Finally, we show that a design paradigm orthogonal to traditional repair mechanisms involving redefinition of the yield criterion by accepting memories with failures is a promising candidate for improving yield without incurring additional overheads.
The paper proposes a new type of linear electrostatic motor operated on a principle of the electric field effect on dielectric material. The proposed motor has a simple structure that simplifies the production process. Moreover, the motor is designed with very light weight rotor. As a result, the friction and power losses of the motor are reduced. Additional advantage of this motor is that it is not influenced by external magnetic fields. Furthermore, the proposed motor has advanced motion control options that allow precise control of velocity and position profiles. Extensive simulation results of the proposed motor at different operation modes are presented and analyzed. The simulation results demonstrate proposed control methods and show the practicability of the motor.
Voice activity detection in the presence of highly non-stationary noise and transient interferences is an open problem. State-of-the-art voice activity detectors which are based on statistical models usually assume that noise is slowly varying with respect to speech. This assumption does not hold for transient interferences which are short time interruptions, and the performance of these detectors significantly deteriorates. In this paper, we propose a supervised learning algorithm for voice activity detection which is designed to perform in the presence of transients. We consider a labeled training set which comprises speech, background noise and transients, and propose a continuous measure for voice activity based on the Support Vector Machine (SVM) classifier. The measure of voice activity is constructed in a features domain, where the features are based on the scattering transform, include noise estimation, and are designed to separate speech and non-speech frames. Experimental results demonstrate that the proposed algorithm outperforms state-of-the-art detectors for different types of background noises, and in particular accurately classifies frames which contain transient interferences.
We develop dynamic equations that describe the dissemination of knowledge in sub-communities, which represent various groups in a community. Knowledge is represented by a set of concepts. With respect to each concept, the community is divided into two parties: the cognoscenti, who have already produced references to the concept, and the dilettanti, who have not yet produced a reference to the concept. The obtained equations take into account the birth-death processes, and determine the dynamic distribution of ages in the sub-communities. In the case of natural community, such equations with the values in the scales of human life, specify real distribution of the ages; however, in the case of social networks and corresponding communities and groups, the birth-death processes stand for the processes of joining the communities, active communication, and leaving the communities or interrupting communication in them. We study the mutual impact of the dynamics of the concepts and groups of concepts on one hand, and the dynamics of the sub-communities on the other hand. We show that the long-term dynamics of the concepts and sub-communities depend on communication resources of the community, capabilities of concepts' comprehension, and distribution of ages. Higher values of these three parameters allow more flexibility in choosing a policy of allocating the communication resource over the set of concepts. Given the three parameters, the community may develop a policy formulated in terms of the informational measure of the set of concepts, as illustrated in the presentation.
This paper presents the design and IC implementation of a fully-digital 10-bit, 4Mbps sampling rate, delay-line analog-to-digital converter (DL-ADC) for power management applications. The design of the ADC is based on the approach of delay cells string to reduce design complexity and the resultant the silicon area. A unique advantage of the new ADC architecture and the design process is that it is entirely based on standard digital cells out of a vendor's library. Namely, neither custom nor analog design is required, making the concept attractive in terms of performance, scalability to other implementation platforms, design complexity and cost. In this study, two implementation options to the DL-ADC architecture are presented, and both are demonstrated and verified with post-layout results on a Tower Jazz 0.18μm power management (TS18PM) platform. The total silicon area that is required for the implementation of the new DL-ADC sums at 0.05mm2, which confirms the area saving attribute of the concept and design procedure.
A new discrete model of the Ricci flow for images is presented, using a purely combinatorial method for calculating the Ricci curvature, based on R. Forman's work on generalized Laplacians for cell complexes. We describe the discrete Ricci curvature operator, and the implementation of the Ricci flow, and then proceed to show the results of applying the flow to both synthetic and real images. Finally, we implement the discrete Ricci flow in some applications of image processing and vision, such as high dynamic range (HDR) imaging, forward-and-backward (FAB) Ricci flow, and change detection in aerial images.
A backlight illuminator with low divergence angle will significantly improve the efficiency, contrast and color gamut of liquid-crystal displays (LCDs) and micro-projectors. However, converting light from a light-emitting diode (LED) with large angle of divergence into a low-angle of divergence beam using a device that is only a few mm thick is extremely challenging task. Here, we propose a novel solution to this problem by using an optimized light-guide collimator and a quasi-diffractive-optical element. The idea is to illuminate the screen with LED through a parabolic waveguide/funnel that converts a point source into a collimated beam with relatively small diameter. The collimated beam at the output of the parabolic collimator is aligned at an angle to a prismatic quassi diffractive optical element (DOE) fabricated on top of a piece of glass. The illumination at an angle increases the size of the illuminating spot and the goal of the DOE is to correct the propagation direction of the illuminating beam. The proposed concept is demonstrated via preliminary numerical and experimental results.
Conformal and their natural generalization to quasi-conformal mappings of surfaces, have been extensively and successfully employed in various tasks of Computer Graphics and Imaging. Due to the intrinsic differences between surfaces and objects from higher dimensions, many important results on surfaces can not be directly generalized to produce desirable mapping of volumes. Moreover, in dimension higher than 2, there are no conformal maps apart from Möbius transformations. Therefore, most of the real-world applications generate only quasi-conformal transformations, which produce some conformal distortion. Hence it is tempting and natural to measure the quality of volume deformation by the amount of a conformal distortion it produces. In this paper we examine theoretical properties of quasi-conformal mappings in 3D. We apply those conclusion to process discrete volumetric data in the terms of conformality. We present numerical methods to measure “the degree of conformality” of a transformation between a given pair of domains, represented by volumetric meshes.
the human brain consists of many different neural elements that comprise a functioning processing and control system. Though each area has its unique role, they do not operate separately. It is empirically known that during brain activity, different elements affect each other. Such neural network's activity is called Functional Brain Connectivity (FBC).It is a known phenomenon among athletes who suffer mild head injuries frequently along their career, to be more likely to suffer from neurological diseases such as Alzheimer's disease or Parkinson's at old age. This phenomenon has been studied by physicians and health researchers as early as 1971 [1]. Patients who suffered from frequent mild head injuries have no neurological appearable symptoms at young age. FBC abnormality accompanies the mentioned neurological diseases, each characterizes with a different pattern [2], [3]. These two facts led the hypothesis that frequent head injuries can, with some probability, cause FBC abnormalities. Proving this hypothesis and mapping the transformed pattern of FBC can contribute to understanding neurological diseases and by this help the search after treatment generally and preventing for frequent mild injured athletes particularly.This work offers a technology that enables brain researchers to evaluate and locate abnormalities in FBC using fMRI data of a patient. It does so using the fact that connectivity pattern, at rest state, is to a large extent symmetrical. The system evaluates connectivity between classes of voxels using a new nonlinear - Differential Filtered Coherence. It is a mathematical solution for describing connections that are not necessary linear by following the influences of changes in one signal on the changes in another.Finally, the system finds FBC asymmetry using registration of the generated connectivity map between the two hemispheres of the brain.
Computer vision is currently one of the most exciting and rapidly evolving fields of science, which affects numerous industries. Research and development breakthroughs, mainly in the field of convolutional neural networks (CNNs), opened the way to unprecedented sensitivity and precision in object detection and recognition tasks. Nevertheless, the findings in recent years on the sensitivity of neural networks to additive noise, light conditions, and to the wholeness of the training dataset, indicate that this technology still lacks the robustness needed for the autonomous robotic industry. In an attempt to bring computer vision algorithms closer to the capabilities of a human operator, the mechanisms of the human visual system was analyzed in this work. Recent studies show that the mechanisms behind the recognition process in the human brain include continuous generation of predictions based on prior knowledge of the world. These predictions enable rapid generation of contextual hypotheses that bias the outcome of the recognition process. This mechanism is especially advantageous in situations of uncertainty, when visual input is ambiguous. In addition, the human visual system continuously updates its knowledge about the world based on the gaps between its prediction and the visual feedback. CNNs are feed forward in nature and lack such top-down contextual attenuation mechanisms. As a result, although they process massive amounts of visual information during their operation, the information is not transformed into knowledge that can be used to generate contextual predictions and improve their performance. In this work, an architecture was designed that aims to integrate the concepts behind the top-down prediction and learning processes of the human visual system with the state-of-the-art bottom-up object recognition models, e.g., deep CNNs. The work focuses on two mechanisms of the human visual system: anticipation-driven perception and reinforcement-driven learning. Imitating these top-down mechanisms, together with the state-of-the-art bottom-up feed-forward algorithms, resulted in an accurate, robust, and continuously improving target recognition model.
Applications of anisotropic diffusion equation to texture enhancement have shown that an image can be smoothed while preserving high frequency features like edges. However, preserving textures still remains challenging. One way to preserve textures is by adding an extra term to the diffusion equation. This additional term can be interpreted as a potential, similar to the role of Schrödinger's potential in the complex diffusion equation, or as a reaction term in a reaction-diffusion process. We show the effect of such potentials on texture denoising, highlighting that anisotropic diffusion with potential combines properties of diffusion (piecewise-smoothing) and potential (enhancing fine structures) filters. Simulations performed on pure texture samples indicate that the reconstruction depends on the type of texture and the transform operator used in the potential. The diffusion-with-potential approach is extended. Local and nonlocal results show that nonlocal diffusion improves the quality of denoising.
Detection of defects on patterned semiconductor wafers is a critical step in wafer production. Many inspection methods and apparatus have been developed for this purpose. We recently presented an anomaly detection approach based on geometric manifold learning techniques. This approach is data-driven, with the separation of the anomaly from the background arising from the intrinsic geometry of the image, revealed through the use of diffusion maps. In this paper, we extend our algorithm to 3D data in multichannel wafer defect detection. We test our algorithm on a set of semiconductor wafers and demonstrate that our multiscale multi-channel algorithm has superior performance when compared to single-scale and single-channel approaches.
